persistent leptin signaling in the arcuate nucleus impairs ... · lating leptin, or...

Original Paper

Neuroendocrinology 2019;109:374–390

Persistent Leptin Signaling in the Arcuate Nucleus Impairs Hypothalamic Insulin Signaling and Glucose Homeostasis in Obese Mice

Eglantine Balland

a Weiyi Chen

a Tony Tiganis

b Michael A. Cowley

a a

Department of Physiology, Metabolism, Diabetes and Obesity Program, Biomedicine Discovery Institute, Monash University, Clayton, VIC, Australia; b Department of Biochemistry and Molecular Biology , Metabolism, Diabetes and Obesity Program, Biomedicine Discovery Institute, Monash University, Clayton, VIC, Australia

Received: February 4, 2019Accepted after revision: April 2, 2019Published online: April 16, 2019

Dr. Eglantine Balland and Prof. Michael A. CowleyDepartment of Physiology, Metabolism, Diabetes and Obesity ProgramBiomedicine Discovery Institute, Monash University26 Innovation Walk, Clayton, VIC 3800 (Australia)E-Mail eglantine.balland @ monash.edu and michael.cowley @ monash.edu

© 2019 S. Karger AG, Basel

E-Mail [email protected]/nen

DOI: 10.1159/000500201

KeywordsDiabetes · Hypothalamus · Insulin resistance · Leptin · Obesity

AbstractBackground: Obesity is associated with reduced physiologi-cal responses to leptin and insulin, leading to the concept of obesity-associated hormonal resistance. Objectives: Here, we demonstrate that contrary to expectations, leptin signaling not only remains functional but also is constantly activated in the arcuate nucleus of the hypothalamus (ARH) neurons of obese mice. This state of persistent response to endogenous leptin underpins the lack of response to exogenous leptin. Methods and Results: The study of combined leptin and insulin signaling demonstrates that there is a common pool of ARH neurons responding to both hormones. More importantly, we show that the constant activation of leptin receptor neurons in the ARH prevents insulin signaling in these neurons, leading to impaired glucose tolerance. Ac-cordingly, antagonising leptin signaling in diet-induced obese (DIO) mice restores insulin signaling in the ARH and improves glucose homeostasis. Direct inhibition of PTP1B in the CNS restores arcuate insulin signaling similarly to leptin

inhibition; this effect is likely to be mediated by AgRP neu-rons since PTP1B deletion specifically in AgRP neurons re-stores glucose and insulin tolerance in DIO mice. Conclu-sions: Finally, our results suggest that the constant activa-tion of arcuate leptin signaling in DIO mice increases PTP1B expression, which exerts an inhibitory effect on insulin sig-naling leading to impaired glucose homeostasis.

© 2019 S. Karger AG, Basel

Introduction

Leptin is produced by adipocytes and instructs the brain on energy stores to regulate energy balance accord-ingly [1, 2]. Circulating leptin levels are proportionate to adiposity level [3]. From the blood, leptin enters the brain [4, 5] and signals via its leptin receptor (LepRb), present in various brain regions and highly expressed within the arcuate nucleus of the hypothalamus (ARH) [6, 7]. LepRb neuronal signaling in the ARH is coordinated via the sig-nal transducer and activator of transcription 3 (STAT3) and protein kinase B/Akt pathways [8] and when acti-vated reduces food intake and body weight [9, 10]. Im-portantly, obesity is associated with a high level of circu-

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M

Persistent Leptin Signaling Impairs Glucose Homeostasis in Obesity

375Neuroendocrinology 2019;109:374–390DOI: 10.1159/000500201

lating leptin, or hyperleptinemia. However, obesity is characterized by the absence of an adequate response to the high level of endogenous leptin [11], this particular state is known as “leptin resistance.” Recent evidence in-dicates that some leptin activity persists in obesity [12, 13]. Leptin has profound effect on body weight regulation and glucose homeostasis [14–17]. Obesity is associated with hyperleptinemia, hyperglycemia, hyperinsulinemia and type 2 diabetes. Besides leptin actions in the ARH, insulin also activates the phospho-inositol-3-kinase/Akt signaling cascade through central insulin receptor (InsR) to maintain normal glucose homeostasis [18]. Insulin re-sistance in the brain occurs in obesity and may be in-volved in type 2 diabetes pathogenesis. Furthermore, re-cent studies also link central insulin resistance with de-pression and neurodegenerative diseases [19, 20].

In the present study, we aim to define the state of “leptin resistance” at the cellular level diet-induced obese (DIO) mice and its possible involvement in glucose ho-meostasis impairment. DIO mice display a constant high basal LepRb signaling in the ARH owing to the presence of hyperleptinemia and this is characterized by sustained activation of both STAT3 and Akt signaling pathways. We hypothesized that the elevated LepRb signaling in DIO mice is associated with attenuated insulin signaling, contributing to the alteration of glucose homeostasis. This study redefines our understanding of central leptin resistance and highlights a previously unexplored mecha-nism by which leptin signaling perturbs the central con-trol of glucose metabolism in diet-induced obesity, defin-ing a possible molecular link between hyperleptinemia-induced neuronal signaling and central insulin resistance.

Materials and Methods

AnimalsAll animal procedures were approved by Monash University

Animal Ethics Committee. Two- to three-month-old male C57BL6 mice (Monash Animal Services) were housed in a controlled envi-ronment with constant temperature and 12 h light/12 h dark cycle in a high-barrier facility (Monash ARL). Mice were either fed ad libitum with standard chow diet (8.5% fat, Barastoc, Ridley AgriProducts, Australia) or high-fat diet (HFD; 43% energy from fat, SF04–001, Specialty Feeds, Western Australia, Australia) for 12–20 weeks. To generate Agrp-Ires-Cre; Ptpn1fl/fl (AgRP-1B), Ptpn1fl/fl (C57BL/6) mice [21] were bred with Agrp-Ires-Cre mice [22].

GenotypingGenotyping was performed by PCR on DNA extracted from

tail biopsies using primers previously described for the Agrp-Ires-Cre [22] and Ptpn1fl/fl mice [21].

Lateral Ventricle Cannulation and Intracerebro-Ventricular InfusionsCannulas were stereotaxically inserted (1.0 mm lateral and 0.2

mm rostral to bregma). Each surgery was followed by 1 week of recovery before performing additional procedures on cannulated mice. For the brain signaling analysis, recombinant murine leptin (1 µg; Peprotech), leptin antagonist (LAN; 5 µg; Protein Laborato-ries Rehovot, Israel), cucurbitacin (5 µg; Sigma-Aldrich), or clara-mine (5 µg; Sigma-Aldrich) were infused 30 min before parafor-maldehyde intracardiac perfusion.

Antibodies used for IHCpSTAT3 (1: 1,000; Cell Signaling Technology #9131, USA)pAkt (1: 1,000; Cell Signaling Technology #4060, USA)γ-MSH (1: 1,000; Antibodies Australia #AS599, Australia)AgRP (1: 1,000; Antibodies Australia #AS506, Australia)

ImmunofluorescenceMice were fasted overnight and injected peripherally in the

morning with saline, with leptin (3 mg/kg; Peprotech) 45 min be-fore perfusion or with insulin (5 units/mouse; Novo Nordisk) 10 min before perfusion. Intracardiac perfusions were performed on isofluorane-anaesthetized mice with 2% paraformaldehyde. Brains were postfixed, cryoprotected overnight at 4 ° C in a 20% sucrose solution, and embedded in Tissue Tek (Miles, Elkhart, IN, USA) before freezing. Thirty micro meter-thick coronal sections were cut on a cryostat and processed for immunofluorescence as previ-ously described [5, 23]. Since pSTAT3 and pAkt antibodies were produced in the same specie (rabbit), we used a sequential staining protocol with a fixation step between the 2 primary antibodies. pSTAT3 antibody was incubated first for 48 h at 4 ° C and subse-quently incubated with a goat-anti-rabbit Alexa 568. After revela-tion of pSTAT3 in red, slices were extensively washed in KPBS and then fixed with 4% PFA for 1 min. Following the fixation step, brain slices were washed in KPBS and incubated with pAkt anti-body for 48 h at 4 ° C. Slices were then revealed using goat-anti-rabbit Alexa 488 green secondary antibody, washed and cover-sliped. Figure 3c illustrates the specificity of our signal and the absence of cross-reactivity that would have generated 100% of co-localization.

Double-immunofluorescence images were acquired using a confocal microscope (Leica SP8 Confocal Invert, Monash Micro-imaging). The number of immunoreactive cells was counted using Image J software in three hemi-sections of similar rostro-caudal level (bregma –1.82 to –2.06) per mouse (see next section for more details).

Immunohistochemistry Images QuantificationMicrophotographs were taken on a confocal microscope with

the same settings. Unmodified pictures (no modification of con-trast, brightest or any other parameter, no threshold applied) were subjected to quantification using Image J software (point tool) on individual channel images. Neurons were considered as specifi-cally labeled when the same researcher was able to clearly identify a colored neuron (green or red) at a fixed distance from the screen (50 cm) on pictures displayed at the same zoom level. Each labeled neuron was marked with a colored dot (green dot on pAkt-labeled neurons placed on green channel picture; red dot on pSTAT3-la-beled neurons placed on red channel picture). After completion of neurons quantification on individual channel picture, the dots

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M

Balland/Chen/Tiganis/CowleyNeuroendocrinology 2019;109:374–390376DOI: 10.1159/000500201

Control DIO

3V

Merged

pAkt

pSTAT3

0

20

40

60

Control DIO

**

Num

ber o

f pAk

t/pS

TAT3

IR c

ells

Basal STAT3/Akt co-activation in ARH

0

102030

21

4050607080

Control Control+ LAN

DIO+ LAN

DIO

***

Num

ber o

f pAk

t/pS

TAT3

IR c

ells

****

db

Control DIOLAN icv

3V

Merged

pAkt

pSTAT3

a c

Fig. 1. DIO mice display high levels of STAT3 and Akt activation in the ARH in basal conditions, which is suppressed by centrally delivered LAN. a Representative photomicrographs (10×) show-ing pSTAT3 (red) and pAkt (green)-labeled cells following ip sa-line injection (scale bar 75 µm). The doted lines represent the re-gion of labeled neurons quantification. b Graph representing the mean number of ARH neurons colabeled for pSTAT3 and pAkt per hemi-section 30 min after saline ip injection. c Representative photomicrographs (10×) showing pSTAT3 labeled cells (red) and pAkt labeled cells (green) 30 min following LAN injection (5 µg) in the LV. STAT3 and Akt basal coactivation is suppressed in con-trol diet mice and DIO mice (scale bar 75 µm). Arrowheads high-

light examples of pSTAT3-alone (red) and pAkt-alone (green) neurons. d Graph representing the mean number of ARH neurons displaying pSTAT3/pAkt colabeled cells per hemi-section in basal conditions or 30 min after LAN icv injection. All data are repre-sented as mean SEM (n = 4 mice/group). ns, p > 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001 as determined by two-tailed unpaired t test (b) and one-way ANOVA followed by Tukey’s post hoc test (d). See also online suppl. Figure S1 and S2. LAN, leptin antago-nist; icv, intracerebro-ventricular; DIO, diet-induced obese; STAT3, signal transducer and activator of transcription; ARH, ar-cuate nucleus of the hypothalamus.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



were pasted on the picture (“flatten” option from the Image tab on Image J). In order to quantify consistently and accurately the co-localizations of pAkt and pSTAT3 staining, individual channels pictures displaying colored dots (pAkt-labeled neurons with green dots and pSTAT3-labeled neurons with red dots) were opened on illustrator software. The transparency level of the pictures was set to 40% and then the green and red channel pictures were super-posed, allowing the quantification of green dots alone (pAkt-alone), red dots alone (pSTAT3-alone), and colocalized green and red dots (pakt/pSTAT3 colocalization).

NB: The same quantification method was used for triple label-ing. The quantification area (ARH) was identified by the detection of the ARH boundaries by the observation of cell nuclei autofluo-rescence or cell nucleus counterstaining using Hoechst. The re-searcher was blind to the treatments and performed a second quantification on several pictures (minimum 4 per experiment) on a different day to ensure the consistency of the quantification. The differences between both quantifications were consistently inferi-or to 5%. Figure 3c and online suppl. Fig. 3 (for all online suppl. material, see www.karger.com/doi/10.1159/000500201).

Ex vivo ExperimentsMice fed with standard chow or HFD for 15–20 weeks were

fasted overnight prior being anesthetized with isofluorane to re-ceive an intracardiac perfusion of cold aCSF. Then brains were quickly extracted and cut in 150 µm slices, using a vibratome. Brain slices were collected from –1.34 to –2.30 mm relative to bregma. After 1 h of recovery during which slices were placed in aCSF oxy-genized with carbogen, they were incubated in fresh aCSF with saline (1 µL) or in fresh aCSF with leptin (100 nM; Peprotech) for 45 min at room temperature. At the end of the treatment, slices were fixed by immersion in 2% PFA and stored at –20 ° C in cryo-protective solution until being proceeded. Immunofluorescence was performed on the slices using the same protocol used for in vivo immunostaining. In this experiment, 6 control and 6 DIO mice were used for the pSTAT3/γ-MSH colabeling and another group of 6 control and 6 DIO mice for the pSTAT3/AgRP cola-beling.

Metabolic MeasurementsGlucose tolerance tests (GTTs) and insulin tolerance tests were

performed on 16 h (Fig. 4e, f), 6 h (Fig. 6h, i), and 4 h (Fig. 6j, k) fasted conscious mice by injecting D-glucose (2 mg/g body weight) or insulin (0.65 mU/g body weight) into the peritoneal cavity and measuring glucose in tail blood immediately before and at 15, 30, 60, 90, and 120 min after intraperitoneal (ip) injection using an Accu-Check glucometer (Roche, Germany). The 7 days of LAN or aCSF injections prior GTT were performed as follow: 5 µg, once daily, 1 h before dark phase. The GTT was performed on day 8, 16 h following the last injection.

Body composition (lean and fat mass) was measured using EchoMRI (Echo Medical Systems, Houston, TX, USA).

StatisticsData were analyzed on GraphPad PRISM and are represented

as mean with error bars indicating SEM. p values were calculated using one-way ANOVA followed by Tukey’s multiple-comparison test, two-way ANOVA followed by Bonferroni’s post hoc test, or t test. ns, p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.

Results

Persistent Leptin Signaling in Arcuate Neurons in DIO MiceAfter 12 weeks on HFD mice had increased body

weight compared to chow fed mice (online suppl. Fig. S1a). When leptin was delivered into the lateral ventricle of control mice (intracerebro-ventricular injection, or icv), we observed an increase in the number of pSTAT3-labeled neurons, but this effect was absent in obese mice (online suppl. Fig. S1b). ip injection of leptin did not in-crease the number of pSTAT3-labeled neurons in the ARH of DIO mice. However, in basal conditions, DIO mice express a higher number of pSTAT3-labeled neu-rons in the ARH compared to lean, chow fed mice (Fig. 1b; online suppl. Fig. S1c), confirming previous findings [23].

We also performed a combined immunostaining for pSTAT3 and pAkt without exogenous leptin stimulation in control and DIO mice. DIO mice displayed an increase in both pSTAT3-labeled and pAkt-labeled neurons in the basal state when compared to control mice. Importantly, most of the pSTAT3-labeled ARH neurons were also pAkt-labeled (Fig. 1a, b). The coactivation of 2 LepRb-associated signaling pathways suggests that endogenous leptin may be responsible for the high basal activation of STAT3 and Akt in the ARH of DIO mice.

To test if hyperleptinemia drives constant LepRb activa-tion in ARH neurons, we injected control and DIO mice icv with LAN [24] to block endogenous leptin from activat-ing LepRb-associated signaling pathways (Fig. 1c). In DIO mice, we observed a striking decrease in the number of neurons that were labeled for both pSTAT3 and pAkt. LAN virtually abolished STAT3 and Akt coactivation in ARH neurons (Fig. 1d). Interestingly, the degree of pSTAT3 and pAkt colocalized expression was similar between con-trol and DIO mice after LAN injection (Fig. 1d; online sup-pl. Fig. S2b). This is consistent with the notion that endog-enous leptin in DIO mice is responsible for most of the STAT3 and Akt coactivation seen in ARH neurons.

Leptin-Induced pSTAT3 in ARH Neurons in Ex vivo Hypothalamic Slices from DIO MiceTo determine whether LepR-expressing neurons in

the ARH of DIO mice remain responsive to leptin, we in-cubated hypothalamic slices from DIO mice ex vivo with vehicle or 100 nM leptin and monitored for leptin signal-ing by staining for pSTAT3. Using this approach, we ab-rogated the hyperleptinemic environment of hypothala-mi in DIO and investigated the response of ARH neurons to exogenous leptin in brain slices taken from lean and

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


Saline Leptin

Cont

rol

DIO

3V

Merged

AgRP

AgRP

pSTAT3

pSTAT3

Cont

rol

Saline Leptin

DIO

3V

Merged

γ-MSH

γ-MSH

pSTAT3

pSTAT3

0

20

40

60

Control+ saline

DIO+ saline

DIO+ leptin

Control+ leptin

STAT

3 ac

tivat

ion

in P

OM

C ne

uron

s, %

**** ***

0

20

40

60

Control+ saline

DIO+ saline

DIO+ leptin

Control+ leptin

STAT

3 ac

tivat

ion

in A

gRP

neur

ons,

% * ***

0

10

20

30

40

50

Control+ saline

Control+ leptin

DIO+ saline

DIO+ leptin

Num

ber o

f lab

eled

neu

rons

p = 0.089 p = 0.077

Total AgRP labeled neuronspSTAT3/AgRP co-labeled neurons

0

10

20

30

40

50

Control+ saline

Control+ leptin

DIO+ saline

DIO+ leptin

Num

ber o

f lab

eled

neu

rons

**

*

Total γ-MSH labeled neuronspSTAT3/γ-MSH co-labeled neurons

b

d

a

c

fe

(For legend see next page.)2

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



DIO mice. Following a 1 h recovery time, 6 whole brain slices (150 µm/slice) containing ARH neurons were ob-tained from each mouse and placed in fresh medium. Of the 6 slices, 3 slices were incubated in aCSF with saline while the other 3 were incubated in aCSF containing leptin (100 nM) for 45 min. Slices were then fixed and im-munohistochemistry for pSTAT3 was used to quantify cellular responses to saline and exogenous leptin in proo-piomelanocortin (POMC) and AgRP neurons (Fig. 2). γ-MSH is a peptide derived from the POMC precursor, less studied than α-MSH, another POMC precursor, well known for its appetite suppressive action. However, γ-MSH offers the advantage to be easily detectable by im-munostaining compared to α-MSH, which is released too rapidly to be observed in the cytoplasm of POMC neu-rons. Therefore, we decided to use γ-MSH as a marker of POMC neurons. Specifically, leptin-activated POMC neurons were stained for pSTAT3 (red) and γ-MSH (green) and the number of pSTAT3-labeled, γ-MSH-labeled, and pSTAT3/γ-MSH-double labeled cells quan-tified. Similarly, AgRP neurons were stained for pSTAT3 (red) and AgRP (green) and the number of pSTAT3-la-beled, AgRP-labeled, and pSTAT3/AgRP-double labeled cells quantified. Interestingly, we found that the level of pSTAT3 did not differ between control and DIO mice under basal conditions in POMC and AgRP neurons, re-spectively (Fig. 2b, d). However, pSTAT3-labeled POMC (Fig. 2a, b) and AgRP (Fig. 2c, d) neurons were elevated similarly in control and DIO mice in the presence of ex-ogenous leptin. These results indicate that ARH POMC and AgRP neurons remain responsive to leptin in DIO. Moreover, these results are in line with the idea that hy-perleptinemia in obesity may account for the high basal pSTAT3 signaling (see also Fig. 2e, f).

Persistent ARH Leptin-Signaling and Diminished Insulin Signaling in DIOSince leptin is capable of coactivating Akt and STAT3

signaling within the same cell, we used it to discriminate between the respective actions of leptin (pAkt/pSTAT3 colabeled cells) and insulin (pAkt-alone labeled cells, not colocalized with pSTAT3) within the ARH (Fig. 1c for il-lustration of pAkt-alone and pAkt/pSTAT3 colabeled neurons). Importantly, exogenous insulin or leptin injec-tion induced a similar increase in the number of pAkt-alone labeled cells and pSTAT3/pAkt colabeled cells in the ARH of lean mice, respectively (Fig. 3a, b, d for sche-matic representation of quantification method). Interest-ingly, the number of pSTAT3/pAkt colabeled and pAkt-alone labeled neurons was significantly increased upon leptin and insulin coinjection compared to saline injec-tion. However, the magnitude of the increase in number of pAkt-alone labeled and pSTAT3/pAkt colabeled neu-rons was markedly attenuated compared to a single injec-tion of either insulin or leptin (Fig. 3a, b). This suggests the existence of a common pool of ARH neurons that re-sponds to both leptin and insulin. In DIO mice, the num-ber of pAkt-alone labeled neurons remained unaltered upon ip insulin injection (Fig. 4a, b; online suppl. Fig. S2a). In line with our earlier findings, DIO mice display a higher number of pSTAT3/pAkt colabeled neurons when compared to age matched control mice after saline injec-tion (Fig. 1a, b, 4a, d), the number of pSTAT3/pAkt-la-beled neurons was also unchanged after insulin injection (Fig. 4a, d; online suppl. Fig. S2a). Noteworthy, hypotha-lamic insulin signaling mostly occurred in the ARH (on-line suppl. Fig. S1d).

In DIO mice, it is evident that a large population of leptin-sensitive neurons in the ARH are responsive to en-dogenous leptin in the basal state and this reduces the ac-tion of exogenous leptin (online suppl. Fig. S1b, S2a) or insulin (Fig. 4a, b; online suppl. Fig. S2a) on these neu-

Fig. 2. Ex vivo slices from DIO mice display leptin-induced pSTAT3. a Representative photomicrographs (20×) of ex vivo slic-es from control and DIO mice showing γ-MSH-labeled cells (green) and pSTAT3-labeled cells (red) following 45 min incuba-tion in aCSF + saline or in aCSF + 100 nM leptin (scale bar 25 µm, merged image). b Graph representing the percentage of pSTAT3 labeling in γ-MSH-expressing ARH neurons following 45 min in-cubation of the slices in aCSF + saline or in aCSF + 100 nM leptin. c Representative photomicrographs (20×) of ex vivo slices from control and DIO mice showing AgRP-labeled cells (green) and pSTAT3-labeled cells (red) following 45 min incubation in aCSF + saline or in aCSF + 100 nM leptin (scale bar 25 µm, merged im-

age). d Graph representing the percentage of pSTAT3 labeling in AgRP-expressing ARH neurons following 45 min incubation of the slices in aCSF + saline or in aCSF + 100 nM leptin. e Graph representing the mean number of cells displaying γ-MSH/pSTAT3 colocalization within the total number of γ-MSH-labeled cells. f Graph representing the mean number of cells displaying AgRP/pSTAT3 colocalization within the mean number of total AgRP-labeled cells. All data are represented as mean SEM (n = 4 control diet and 6 DIO mice). * p < 0.05; *** p < 0.001; **** p < 0.0001 as determined by one-way ANOVA followed by Tukey’s post hoc test (b, d–f). DIO, diet-induced obese; STAT3, signal transducer and activator of transcription; POMC, proopiomelanocortin.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


Saline Leptin Leptin + insulinInsulin

pAkt

Hoechst

3V

pSTAT3

3V

pAkt alone

pAkt/pSTAT3pSTAT3 alone

pAkt alone

pSTAT3 alone

pAkt/pSTAT3

0

20

40

60

80

100

120

Num

ber o

f IR

cells

Saline Insulin Leptin Leptin + insulin

****

****

**

****

****

***

**Saline

Leptin Leptin + insulin

Total = 47 Total = 123

Total = 114 Total = 121

Insulin

23

9728

713

94

9

31

81

109

3

b

a

c d

3(For legend see next page.)

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



pAkt

pSTAT3

pAkt

pSTAT3

pAkt

pSTAT3

pAkt

pSTAT3Saline Insulin

DIO

Cont

rol

Merged

3V

3V

Merged

a

Fig. 4. Centrally delivered LAN restores insulin-induced Akt acti-vation in the ARH of DIO mice. a Representative photomicro-graphs (10×) showing pSTAT3 (red) and pAkt (green)-labeled cells 10 min following saline or insulin ip injection (5 U/mouse) in control and DIO mice (scale bar 75 µm, merged image). b Graph representing the mean number of ARH neurons displaying pAkt-alone labeling (pAkt-positive/pSTAT3-negative cells) per hemi-section. c Representative micrograph (10×) of pSTAT3-labeled (red) and pAkt-labeled (green) neurons after central delivery of LAN followed by insulin ip injection in DIO mice (scale bar 75 µm, merged image). d Graph representing the mean number of ARH

neurons displaying pAkt/pSTAT3 colabeling per hemi-section. e GTT in control and DIO mice following 7 consecutive days of aCSF or LAN (5 mg once daily) icv injections. f Area under the curve of the GTT presented in (e). All data are represented as mean SEM. ns, p > 0.05; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001 as one-way ANOVA followed by Tukey’s post hoc test (b, c: n = 4–8 mice/group; e, f: n = 5/7 mice/group). See also online supplementary Figure S2. LAN, leptin antagonist; DIO, diet-in-duced obese; STAT3, signal transducer and activator of transcrip-tion; ARH, arcuate nucleus of the hypothalamus.

(Figure continued on next page.)

Fig. 3. Leptin and insulin activate the same pool of ARH neurons. a Representative photomicrographs (20×) showing pSTAT3-la-beled cells (red) and pAkt-labeled cells (green) following saline, insulin (10 min), leptin (45 min) or leptin + insulin ip injection in lean mice (scale bar 75 μm). b Graphs and pie charts representing the mean number of arcuate neurons displaying pSTAT3-alone, pAkt-alone, and colocalized pSTAT3/pAkt labeling per hemi-sec-tion after saline, insulin, leptin and leptin + insulin ip injection. c Representative photomicrographs (20×) showing pSTAT3 (red) and pAkt (green) labeled cells to illustrate neurons expressing pAkt-alone (green arrows), pSTAT3-alone (red arrows) or colo-calized pSTAT3/pAkt (white arrows). d Schematic representing

the method used for neurons quantification. The top green rect-angle illustrates pAkt-labeled neurons quantification on single channel image (green), the top red rectangle illustrates pSTAT3-labeled neurons quantification on single channel image (red), and the green and red rectangles in the bottom illustrate the superposi-tion of both single channel images, with their respective quantifi-cation dots pasted in the image, used to quantify pAkt/pSTAT3 colocalizations. Data are represented as mean ± SEM. ** p < 0.01; *** p < 0.001; **** p < 0.0001 as determined by one-way ANOVA followed by Tukey’s post hoc test (n = 4 mice per group). See also online supplementary Figures S2 and S3.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


rons. Accordingly, we tested if we could improve the neu-ronal action of insulin by injecting LAN icv before giving insulin peripherally. Strikingly, the level of insulin-in-duced Akt activation in DIO mice was restored to a simi-lar level observed in control mice (Fig. 4b, c; online suppl. Fig. S2c). This suggests that blocking central leptin signal-ing improves insulin’s capacity to activate Akt in ARH neurons of DIO mice. To investigate whether the rescue

of insulin signaling by LAN in DIO mice has a physiolog-ical effect on whole-body glucose homeostasis, we per-formed a GTT following 7 days of icv injections of either ACSF or LAN. In DIO mice, LAN treatment significantly improved glucose tolerance compared to mice given ACSF, proving that central inhibition of leptin signaling was able to normalize the response to glucose challenge in DIO mice (Fig. 4e, f).

pAkt

pSTAT3DIO + LAN + insulin

Merged

3V

0

20

40

60

80

100

Num

ber o

f pAk

t IR

cells

*** ***

ns

pAkt alone

CTL + sa

line

CTL + in

sulin

DIO + sa

line

DIO + in

sulin

DIO + LA

N

+ insu

lin

b c

0

20

40

60

80

100

Num

ber o

f pAk

t/pS

TAT3

IR c

ells

**

****

**

STAT3/Akt co-activation in ARH

CTL + sa

line

CTL + in

sulin

DIO + sa

line

DIO + in

sulin

DIO + LA

N

+ insu

lin

ns

0

20

40

60

80

100

Area

und

er th

e cu

rve

CTL+ ACSF

CTL+ LAN

DIO+ ACSF

DIO+ LAN

* *

ns

0

5

10

15

20

25

Glyc

emia

, mm

ol/L

0 30 60Time, min

90 120

CTL + ACSF

DIO + ACSFCTL + LAN

DIO + LAN

d

e

f

4D

ownl

oade

d by

: M

onas

h U

nive

rsity

130.

194.

146.

140

- 11

/29/

2019

4:5

2:00

AM



pAkt pSTAT3

Cucurbitacin/saline Cucurbitacin/insulinDIO

pAkt

Claramine/saline Claramine/insulin

pSTAT3

0

20

40

60

80

Num

ber o

f pAk

t IR

cells

InsulinSaline

*

icv cucurbitacinpAkt alone

0

20

40

60

80

Num

ber o

f pST

AT3/

pAkt

co-lo

caliz

ed IR

cel

ls

InsulinSaline

icv cucurbitacinpSTAT3/pAkt

ba

d

c

0

20

40

60

100

80

Num

ber o

f pAk

t IR

cells

InsulinSaline

icv claramine

pAkt alone

**

0

20

40

60

100

80

Num

ber o

f pST

AT3/

pAkt

co-lo

caliz

ed IR

cel

ls

InsulinSaline

icv claramine

pSTAT3/pAkt

0

20

40

60

100

80

Num

ber o

f pST

AT3

IR c

ells

InsulinSaline

icv claramine

pSTAT3 alone

e f g

Fig. 5. Central inhibition of JAK/STAT or PTP1B restores insulin-induced Akt activation in the ARH of DIO mice. a Representative photomicrographs (20×) showing pSTAT3-labeled cells (red) and pAkt-labeled cells (green) after central delivery of cucurbitacin (5 μg/mouse) followed by saline or insulin ip injection (5 U/mouse) in DIO mice (scale bar 75 µm). b Graph representing the mean number of arcuate neurons displaying pAkt-alone labeling (pAkt-labeled/pSTAT3-negative cells) per hemi-section. c Graph repre-senting the mean number of arcuate neurons displaying double pAkt/pSTAT3 labeling per hemi-section. d Representative photo-micrographs (20×) showing pSTAT3-labeled cells (red) and pAkt-labeled cells (green) after central delivery of claramine (5 µg/

mouse) followed by saline or insulin ip injection (5 U/mouse) in DIO mice (scale bar 75 µm, merged image). e Graph representing the mean number of ARH neurons displaying pAkt-labeling alone (pAkt-positive/pSTAT3-negative cells) per hemi-section. f Graph representing the mean number of ARH neurons displaying dou-ble-labeling for pAkt and pSTAT3 per hemi-section. g Graph rep-resenting the mean number of ARH neurons displaying pSTAT3-labeling alone (pSTAT3-positive/pAkt-negative cells) per hemi-section. All data are represented as mean ± SEM. ns, p > 0.05; * p < 0.05; ** p < 0.01 as determined by two-tailed unpaired t test (b, c: n = 3–4 mice/group; e–g: n = 3 mice/group). icv, intracere-bro-ventricular; DIO, diet-induced obese.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


Next, we used cucurbitacin (5 µg/mouse icv) to inhib-it JAK/STAT pathway [25]; it reduced the number of pSTAT3/pAkt colabeled neurons to less than 20 cells per hemi-section in the ARH (Fig. 5a–c). When insulin was administered peripherally following LAN or cucurbitacin icv injection, insulin-induced Akt activation was ∼2-fold higher in the ARH of DIO mice and similar to the insulin response observed in control mice (Fig. 4a, b, 5a, b and online suppl. Fig. S2c). These results are consistent with obesity driving increased neuronal LepRb signaling in the ARH to alter neuronal insulin signaling in ARH neurons, resulting in glucose homeostasis alteration.

Pharmacological Inhibition of PTP1B in the CNS Restores Hypothalamic Insulin Signaling Similarly to LepR Antagonism in DIO MicePTP1B is part of a feedback loop that negatively regu-

lates leptin and insulin signaling upon receptor activation [21, 26–30]. PTP1B is known to be elevated in the hypo-thalamus in obesity [28, 31–33]. Since leptin can promote PTP1B expression in response to LepR activation [34], we hypothesized that the persistent activation of Leptin sig-naling in obesity is associated to a greater PTP1B-medi-ated inhibition of insulin signaling in the ARH. To test this hypothesis, we investigated in vivo leptin and insulin signaling using immunohistochemistry in conjunction with the icv administration of the PTP1B inhibitor clara-mine [35]. The number of pAkt-alone labeled neurons in response to insulin (ip) was increased after central PTP1B inhibition (claramine icv) in DIO mice (Fig. 5d–g).

These results suggest that the attenuation of insulin signaling caused by the sustained ARH LepRb signaling is mediated through the induction of PTP1B in DIO mice.

PTP1B Inhibitory Effect on Insulin Signaling is Mediated by AgRP NeuronsAgRP and POMC neurons are 2 major ARH neuronal

populations that express high levels of LepRb and InsR. We used triple immunofluorescent labeling to assess leptin and insulin signaling in these neurons. We ob-served a significant increase in pAkt in AgRP neurons following insulin ip injection (Fig. 6a, c). We found that leptin increased the coactivation of Akt and STAT3 (Fig. 6a, c). Strikingly, the significant increase in Akt-alone activation in response to insulin was lost when leptin was coinjected with insulin while the coactivation of Akt and STAT3 in AgRP neurons remained signifi-cantly elevated (Fig. 6a, c). In POMC neurons (marked by γ-MSH labeling), insulin significantly increased Akt-alone activation (Fig. 6b, d). Similarly, leptin increased the coactivation of Akt and STAT3 in POMC neurons (Fig. 6b, d). In contrast to AgRP neurons, both leptin-induced pAkt/pSTAT3 and insulin-induced pAkt-alone remained significantly elevated in POMC neurons when leptin and insulin were coinjected (Fig. 6b, d), highlight-ing the distinct responses of these types of ARH neurons.

Since leptin seems to influence the insulin response primarily in AgRP neurons, we bred Ptpn1fl/fl (C57Bl/6) mice [21] with Agrp-Ires-Cre mice [22]. The resulting mice failed to express PTP1B specifically in AgRP neu-rons (AgRP-1B mice). AgRP-1B mice developed diet-in-

Fig. 6. Acute deletion of PTP1B in the AgRP neurons increases in-sulin sensitivity and ameliorates glucose homeostasis. a Represen-tative photomicrographs (10×) showing pAkt-labeled cells (green), pSTAT3-labeled cells (red), and AgRP-labeled cells (magenta) fol-lowing leptin (45 min) or insulin (10 min) or leptin and insulin ip injection in lean mice (scale bar 75 µm, merged image). b Repre-sentative photomicrographs (10×) showing pAkt-labeled cells (green), pSTAT3-labeled cells (red), and γ-MSH-labeled cells (ma-genta) following leptin (45 min) or insulin (10 min) or leptin and insulin ip injection in lean mice (scale bar 75 µm, merged image). c Graphs representing the mean number of ARH neurons per he-mi-section displaying pSTAT3-labeling in AgRP-labeled neurons, pAkt-labeling in AgRP-labeled neurons, colocalized pSTAT3/pAkt labeling in AgRP-labeled neurons and total number of AgRP-labeled neurons after saline, insulin, leptin and leptin + insulin ip injection. d Graphs representing the mean number of ARH neu-rons per hemi-section displaying pSTAT3-labeling in γ-MSH-labeled neurons, pAkt-labeling in γ-MSH-labeled neurons, colo-calized pSTAT3/pAkt labeling in γ-MSH-labeled neurons, and to-

tal number of γ-MSH-labeled neurons after saline, insulin, leptin, and leptin + insulin ip injection. e Mean body weight of Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding (n = 9–12 mice per group). f Mean body fat mass of Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding (n = 9–12 mice per group). g Mean body lean mass of Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding (n = 9–12 mice per group). h GTT in Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding. i Graph showing the mean area under the curve of GTT in Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feed-ing. j ITT in Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding. k Graph showing the mean area under the curve of ITT in Ptpn1fl/fl and AgRP-1B mice following 16 weeks of high-fat feeding. All data are represented as mean ± SEM. ns, p > 0.05; * p < 0.05; *** p < 0.001; **** p < 0.0001 as determined by one-way ANOVA followed by Tukey’s post hoc test (c, d: n = 4 mice/group) and two-tailed unpaired t test (e–g, i, k: n = 9–12 mice/group). GTT, glucose tolerance test; ITT, insulin tolerance test.

(For figure see next page.)

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



pAkt

AgRP

pSTAT3

Saline Insulin


pAkt

γ-MSH

pSTAT3

Saline Insulin


0102030405060

Body

wei

ght,

g

Ptpn1fI/fI AgRP-1B

p = 0.069

05

1015202530

Fat m

ass,

g

Ptpn1fI/fI AgRP-1B05

1015202530

Lean

mas

s, g

Ptpn1fI/fI AgRP-1B

0

20

40

60

80

Num

ber o

f lab

elle

d ce

lls

Saline Insulin Leptin Leptin+ insulin

**** **

**

***

*

***

ns

pAkt/AgRP

pSTAT3/AgRP

pAkt/pSTAT3/AgRP

Total AgRP

0

20

40

60

80

Num

ber o

f lab

elle

d ce

lls

Saline Insulin Leptin Leptin+ insulin

**** ****

*** **

*

****

pAkt/MSH

pSTAT3/MSH

pAkt/pSTAT3/MSH

Total MSH

0

50

100

150

Area

und

edr t

he c

urve

Ptpn1fI/fI AgRP-1B

GTT

*

0

20

40

60

Area

und

edr t

he c

urve

ITT

p = 0.05

4

6

8

10

12

Bloo

d gl

ucos

e, m

mol

/L

0 30 60 90 120Time, min

Ptpn1fl/fl AgRP-1B

0

10

20

30

40

Bloo

d gl

ucos

e, m

mol

/L

0 30 60 90 120Time, min

Ptpn1fl/fl AgRP-1B

a b

c

e

h i j k

f g

d

6

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


duced obesity when fed with HFD during 16 weeks (Fig. 6e). Body weight, fat mass, and lean mass were sim-ilar between AgRP-1B mice and Ptpn1fl/fl littermates (Fig. 6e–g). Noteworthy, despite developing diet-in-duced obesity, AgRP-1B mice displayed a greater toler-ance to glucose (Fig. 6h, i) and an improved insulin sen-sitivity compared to Ptpn1fl/fl mice (Fig. 6j, k). These findings are consistent with leptin signaling inducing PTP1B-mediated repression of insulin signaling in AgRP neurons, resulting in the alteration of glucose homeosta-sis in obesity.

Discussion

In the current study, we sought to address the hor-monal resistance in the hypothalamus associated with obesity. Our results demonstrate an unexplored causal link between leptin and insulin resistance in the hypo-thalamus of DIO mice. The constant activation of leptin signaling in ARH is responsible, at least in part, for the alteration of hypothalamic insulin signaling and its asso-ciated regulation of glucose homeostasis.

In the present study, mice were fed with HFD for a pe-riod of 15–20 weeks. We first validated the absence of ac-tivation of leptin signaling pathways in response to exog-enous leptin administration; a hallmark of cellular leptin resistance, characterized by the absence of leptin-induced increase in pSTAT3-labeled neurons in the ARH. The lack of cellular and physiological response to leptin in DIO mice cannot be explained by an impairment of leptin transport from the blood to the brain. After 8 weeks on HFD, mice do not respond to peripherally injected leptin but the pSTAT3 response to centrally-injected leptin re-mains intact [36, 37]. Leptin transport can be pharmaco-logically rescued in 8 weeks DIO mice, allowing the res-toration of leptin-induced STAT3 activation in ARH neurons to a level similar to control mice [5]. However, when mice are fed with HFD beyond 12 weeks, there is neither a cellular nor a physiological response to exoge-nous leptin, even if the hormone is directly injected into the brain.

We observed a high level of STAT3 activation under basal conditions concomitant with resistance to exoge-nous leptin administration. This event was characterized by an absence of any further increase in the number of pSTAT3-labeled neurons of DIO mice in response to leptin injection compared to basal conditions with no change in body weight, as previously reported [38]. More importantly, our results consistently showed that ARH

neurons remained responsive to endogenous leptin. We demonstrated ARH neuronal response to endogenous leptin in DIO by using a competitive LAN in vivo. We also demonstrated an intact cellular response to exoge-nous leptin treatment in DIO mice using ex vivo hypo-thalamic slices, a condition that negated the hyperleptin-emic environment in DIO mice. Altogether, our results suggest that the response to endogenous leptin remains intact in DIO mice.

Transgenic mice expressing a constitutively active form of STAT3 become obese on HFD [39]. Therefore, constitutive activation of LepRb signaling pathways could be responsible for the lack of physiological effect of exog-enous leptin in DIO mice. In this context, understanding the mechanisms of this phenomenon is crucial. The con-stitutive activation of STAT3 signaling pathway in the ARH neurons of DIO mice has been reported previously [24, 40]. Here we replicated this result but more impor-tantly we demonstrate that DIO mice not only display a constitutive activation of STAT3 signaling but also Akt. Moreover, the activated form of these LepRb-associated signaling molecules colocalized in ARH neurons of DIO mice, and are suppressed by LAN icv injection, suggest-ing that endogenous leptin itself could be responsible for the high colocalized expression of pSTAT3 and pAkt ob-served in basal conditions.

Consistent with this notion, Friedman’s group ele-gantly showed that leptin-deficient (ob/ob) mice fed a HFD and receiving continuous infusion of leptin to match lean-like leptin levels, remained highly sensitive to leptin [41]. This study suggested that hyperleptinemia, rather than excess adiposity or dietary fats, is necessary to blunt the response to exogenous leptin.

Recently, Ottaway et al. [12] demonstrated that endog-enous leptin remains active in DIO mice to regulate food intake and body weight. They show that the administra-tion of a long lasting LAN (PEG-LAN) daily for 7 days induced a similar increase in food intake and body weight in control diet and HFD fed mice. This experiment dem-onstrates that the effect of endogenous leptin, despite be-ing attenuated (the effect is not as high as the level of en-dogenous leptin would predict), is still present in DIO animals. However, such effect was absent in leptin defi-cient (ob/ob) and leptin-receptor deficient (db/db) mice [12]. In our study, we decided to use a nonpegylated LAN to observe the effect of leptin signaling inhibition on in-sulin response and glucose homeostasis, in absence of body weight change.

Leptin and insulin are 2 hormones with many com-mon features; both are released into the blood and they

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



share neuronal targets in the brain [42]. A recent study demonstrated that 50% of arcuate LepR-expressing neu-rons at least are also responsive to central administration of insulin [43]. Their central actions are deeply involved in the regulation of both energy balance [9, 18, 44, 45] and glucose homeostasis [17, 46–49]. Our data indicate that leptin and insulin can act on the same neuronal popula-tion in the ARH and that the sustained activation of lep-Rb-dependent signaling pathways impairs insulin action on the same neurons in the ARH of DIO mice.

Hypothalamic insulin resistance also occurs in obese humans [50, 51]. The inhibition of central InsR signaling with a competitive insulin-receptor antagonist results in hyperglycemia and glucose intolerance in rats [52], sug-gesting that the lack of arcuate insulin signaling is in-volved in the perturbation of systemic glucose homeosta-sis associated with obesity. We thus build on the premise that the constitutive activation of leptin neuronal targets in DIO mice due to hyperleptinemia prevents further in-sulin-induced activation of the same ARH neurons in-volved in glucose homeostasis.

The activation of lepR and insR respectively leads to the activation of a common intracellular signaling cas-cade, the phospho-inositol-3-kinase/protein kinase B (Akt) pathway, leading to the phosphorylation of Akt on 2 sites: T308 and S473 [53, 54]. Since LepR activation also leads to the phosphorylation of STAT3 [54], we used this dual activation of pSTAT3 and pAkt exerted by LepR to set up a novel immunohistochemistry quantification technique, able to discriminate between leptin-respon-sive neurons (Akt/STAT3 coactivation) and the insulin-responsive neurons (activation of Akt only).

Using this technique, we show that antagonizing Lep-Rb is capable of rescuing insulin-induced signaling in the ARH of DIO mice. Then, we examined the effect of the rescue of insulin signaling in the ARH on the regulation of glucose homeostasis and observed that DIO mice treat-ed with LAN display a significantly improved tolerance to glucose. These results are consistent with a recent study where we used another paradigm to investigate the effect lepR antagonism on glucose homeostasis. Importantly, we show that the central inhibition of leptin signaling ex-erted by LAN resulted in the rescue of hepatic glucose production during hyperinsulinemic-euglycemic clamp in DIO mice [63].

Although some studies suggest that insulin transport into the brain is reduced in obese mice [55] and dogs [56], our results suggest that insulin transport persists, at least in part, in DIO mice. In fact, we observe an increase in pAkt levels upon intraperitoneal insulin injection, which

demonstrated that exogenous insulin can reach the ARH. However, further investigation is necessary to define the mechanisms of insulin transport.

The constant high level of leptin in the ARH of obese mice is associated with an increased expression of signal-ing inhibitors such as SOCS3 [57, 58], PTP1B [29, 30, 59], and TCPTP [60]. In standard conditions, these negative regulators are part of a feedback mechanism in response to the activation of the LepRb-dependent signaling cas-cade. Accordingly, SOCS3 mRNA levels were significant-ly lower in DIO mice chronically treated with the LAN [12] and PTP1B or TCPTP neuronal deletion enhances leptin sensitivity [21, 60]. However, the persistent activa-tion of LepRb in response to hyperleptinemia makes these feedback mechanisms inefficient in significantly reduc-ing pSTAT3/pAkt levels.

Our data suggest the existence of a cross-talk between leptin and insulin signaling in the ARH. This notion is supported by evidence showing that SOCS3 deletion in LepRb-expressing neurons protects mice from diet-in-duced insulin resistance independently of body weight [61]. In our study, the persistent activation of lepR in obe-sity concurrently impairs insulin signaling in neurons that are involved in glucose homeostasis. First, we show that antagonizing LepRb restores insulin-induced signal-ing in the ARH of DIO mice and is associated with im-proved glucose tolerance. Then to determine how an in-crease in LepRb signaling may contribute to hypothalam-ic insulin resistance in DIO mice, we specifically targeted PTP1B as a potential mediator of this effect. PTP1B ex-pression is induced by leptin signaling [30], and it is a potent negative regulator of insulin signaling, dephos-phorylating the InsR and InsR substrate-1 [26–29]. PTP- 1B also inhibits leptin signaling [21, 28, 30], the genetic ablation of PTP1B in leptin-responsive neurons results in hypersensitivity to leptin and resistance to diet-induced obesity [62]. Our recent study showed that the elevated level of PTP1B expression in the mediobasal hypothala-mus of DIO mice was normalized after 3 days of LAN icv injections [63]. In the present study, central inhibition of PTP1B restores insulin-induced Akt activation in the ARH of DIO mice, similarly to central blockade of leptin signaling. Our observation is supported by the phenotype of the mice lacking PTP1B specifically in neurons. Note-worthy, these animals display increased insulin sensitiv-ity and improved glucose tolerance even on HFD [21].

Since leptin and insulin can act on the same arcuate neuronal populations [43, 45], we hypothesize that per-sistent leptin signaling in DIO mice could result in en-hanced PTP1B activity to blunt insulin signaling in the

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


ARH neurons involved in glucose homeostasis. We in-vestigated leptin and insulin signaling in the 2 major pop-ulations of ARH neurons regulating glucose homeostasis, namely, POMC and AgRP neurons. Our immunohisto-chemistry studies reveal the convergence of leptin and insulin signaling in AgRP neurons. However, we observe 2 distinctive subpopulations of POMC neurons, which preferentially respond to insulin versus leptin respective-ly when both hormones are coadministrated. This is con-sistent with previous findings including the recent single-cell sequencing of POMC neurons, showing the hetero-geneity of expression of lepR and InsR in this neuronal population [45, 64, 65]. On the contrary, the heteroge-neous response to leptin and insulin did not exist in AgRP neurons. Specifically, the phosphorylation of Akt alone was blunted when leptin and insulin were coinjected while the colocalization of pAkt/pSTAT3 remained ele-vated. This corresponds to the loss of insulin signaling in AgRP neurons in a context of elevated leptin levels. Fur-thermore, leptin per se reduces the number of AgRP neu-rons expressing pAkt-alone, pointing toward a reduced response to endogenous insulin. Overall, these results in-dicate that prolonged activation of leptin signaling may exert an inhibitory effect via PTP1B on insulin signaling in AgRP neurons, but not in POMC neurons.

Building on this notion, we generated mice deletion of PTP1B specifically in AgRP neurons (AgRP-1B mice). In-terestingly, AgRP-1B mice displayed improved glucose tolerance and insulin sensitivity despite their diet-in-duced obesity, in accordance with our in vivo pharmaco-logical inhibition. This point toward the role of AgRP neurons in the development of hypothalamic insulin re-sistance associated to obesity and the resulting hypergly-cemia.

In conclusion, our findings indicate that the “leptin resistant” state (i.e., no appropriate physiological re-sponse) in obese mice is characterized by a persistent re-sponse to endogenous leptin at the cellular level. Obesity-induced hyperleptinemia causes the activation of leptin

signaling in the ARH neurons, which impairs insulin sig-naling in AgRP neurons through the induction of PTP1B. Our findings demonstrate that this state of persistent leptin signaling in the ARH underpins the development of hypothalamic insulin resistance, and it accounts, at least in part, for the perturbation of glucose homeostasis in obesity.

Acknowledgments

The authors acknowledge the facilities, scientific, and technical assistance of Monash Micro Imaging, Monash University, Victo-ria, Australia.

Statement of Ethics

Animal experiments conform to internationally accepted stan-dards and have been approved by the appropriate institutional re-view body.

Disclosure Statement

The authors have no conflicts of interest to declare.

Funding Sources

This research was supported by a Postdoctoral Fellowship to E.B. from Fondation pour la Recherche Medicale (FRM SPE20140129012), Diabetes Australia Grant to E.B. and M.A.C. (DART Grant 2016). NHMRC Fellowships 1079422 to M.A.C. and 1103037 to T.T., NHMRC Grants 1100240 to M.A.C. and T.T. and Grants 1065641, 1103193 to M.A.C.

Authors Contributions

E.B.: designed and performed the experiments. E.B. and W.C.: wrote the manuscript. M.A.C. and T.T.: supervised the work and edited the manuscript.

References 1 Campfield LA, Smith FJ, Guisez Y, Devos R, Burn P. Recombinant mouse OB protein: evi-dence for a peripheral signal linking adiposity and central neural networks. Science. 1995 Jul; 269(5223): 546–9.

2 Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse obese gene and its human ho-mologue. Nature. 1994 Dec; 372(6505): 425–32.

3 Maffei M, Halaas J, Ravussin E, Pratley RE, Lee GH, Zhang Y, et al. Leptin levels in human and rodent: measurement of plasma leptin and ob RNA in obese and weight-reduced subjects. Nat Med. 1995 Nov; 1(11): 1155–61.

4 Banks WA, Kastin AJ, Huang W, Jaspan JB, Maness LM. Leptin enters the brain by a satu-rable system independent of insulin. Peptides. 1996; 17(2): 305–11.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M



5 Balland E, Dam J, Langlet F, Caron E, Stecu-lorum S, Messina A, et al. Hypothalamic tany-cytes are an ERK-gated conduit for leptin into the brain. Cell Metab. 2014 Feb; 19(2): 293–301.

6 Tartaglia LA, Dembski M, Weng X, Deng N, Culpepper J, Devos R, et al. Identification and expression cloning of a leptin receptor, OB-R. Cell. 1995 Dec; 83(7): 1263–71.

7 Scott MM, Lachey JL, Sternson SM, Lee CE, Elias CF, Friedman JM, et al. Leptin targets in the mouse brain. J Comp Neurol. 2009 Jun;

514(5): 518–32. 8 Robertson SA, Leinninger GM, Myers MG Jr.

Molecular and neural mediators of leptin ac-tion. Physiol Behav. 2008 Aug; 94(5): 637–42.

9 Halaas JL, Gajiwala KS, Maffei M, Cohen SL, Chait BT, Rabinowitz D, et al. Weight-reduc-ing effects of the plasma protein encoded by the obese gene. Science. 1995 Jul; 269(5223):

543–6.10 Friedman JM, Halaas JL. Leptin and the regu-

lation of body weight in mammals. Nature. 1998 Oct; 395(6704): 763–70.

11 Frederich RC, Hamann A, Anderson S, Löll-mann B, Lowell BB, Flier JS. Leptin levels re-flect body lipid content in mice: evidence for diet-induced resistance to leptin action. Nat Med. 1995 Dec; 1(12): 1311–4.

12 Ottaway N, Mahbod P, Rivero B, Norman LA, Gertler A, D’Alessio DA, et al. Diet-induced obese mice retain endogenous leptin action. Cell Metab. 2015 Jun; 21(6): 877–82.

13 Simonds SE, Pryor JT, Ravussin E, Greenway FL, Dileone R, Allen AM, et al. Leptin medi-ates the increase in blood pressure associated with obesity. Cell. 2014 Dec; 159(6): 1404–16.

14 Elmquist JK, Coppari R, Balthasar N, Ichinose M, Lowell BB. Identifying hypothalamic pathways controlling food intake, body weight, and glucose homeostasis. J Comp Neurol. 2005 Dec; 493(1): 63–71.

15 Gutiérrez-Juárez R, Obici S, Rossetti L. Mela-nocortin-independent effects of leptin on he-patic glucose fluxes. J Biol Chem. 2004 Nov;

279(48): 49704–15.16 Coppari R, Ichinose M, Lee CE, Pullen AE,

Kenny CD, McGovern RA, et al. The hypo-thalamic arcuate nucleus: a key site for medi-ating leptin’s effects on glucose homeostasis and locomotor activity. Cell Metab. 2005 Jan;

1(1): 63–72.17 Berglund ED, Vianna CR, Donato J Jr, Kim

MH, Chuang JC, Lee CE, et al. Direct leptin action on POMC neurons regulates glucose homeostasis and hepatic insulin sensitivity in mice. J Clin Invest. 2012 Mar; 122(3): 1000–9.

18 Brüning JC, Gautam D, Burks DJ, Gillette J, Schubert M, Orban PC, et al. Role of brain insulin receptor in control of body weight and reproduction. Science. 2000 Sep; 289(5487):

2122–5.19 Moulton CD, Pickup JC, Ismail K. The link

between depression and diabetes: the search for shared mechanisms. Lancet Diabetes En-docrinol. 2015 Jun; 3(6): 461–71.

20 Spielman LJ, Little JP, Klegeris A. Inflamma-tion and insulin/IGF-1 resistance as the pos-sible link between obesity and neurodegen-eration. J Neuroimmunol. 2014 Aug; 273(1–2): 8–21.

21 Bence KK, Delibegovic M, Xue B, Gorgun CZ, Hotamisligil GS, Neel BG, et al. Neuronal PT-P1B regulates body weight, adiposity and leptin action. Nat Med. 2006 Aug; 12(8): 917–24.

22 Tong Q, Ye CP, Jones JE, Elmquist JK, Lowell BB. Synaptic release of GABA by AgRP neu-rons is required for normal regulation of en-ergy balance. Nat Neurosci. 2008 Sep; 11(9):

998–1000.23 Balland E, Cowley MA. Short-term high-fat

diet increases the presence of astrocytes in the hypothalamus of C57BL6 mice without alter-ing leptin sensitivity. J Neuroendocrinol. 2017 Oct; 29(10):e12504.

24 Enriori PJ, Sinnayah P, Simonds SE, Garcia Rudaz C, Cowley MA. Leptin action in the dorsomedial hypothalamus increases sympa-thetic tone to brown adipose tissue in spite of systemic leptin resistance. J Neurosci. 2011 Aug; 31(34): 12189–97.

25 Boykin C, Zhang G, Chen YH, Zhang RW, Fan XE, Yang WM, et al. Cucurbitacin IIa: a novel class of anti-cancer drug inducing non-reversible actin aggregation and inhibiting survivin independent of JAK2/STAT3 phos-phorylation. Br J Cancer. 2011 Mar; 104(5):

781–9.26 Galic S, Hauser C, Kahn BB, Haj FG, Neel BG,

Tonks NK, et al. Coordinated regulation of insulin signaling by the protein tyrosine phos-phatases PTP1B and TCPTP. Mol Cell Biol. 2005 Jan; 25(2): 819–29.

27 Goldstein BJ, Ahmad F, Ding W, Li PM, Zhang WR. Regulation of the insulin signal-ling pathway by cellular protein-tyrosine phosphatases. Mol Cell Biochem. 1998 May;

182(1-2): 91–9.28 Zhang ZY, Dodd GT, Tiganis T. Protein Ty-

rosine Phosphatases in Hypothalamic Insulin and Leptin Signaling. Trends Pharmacol Sci. 2015 Oct; 36(10): 661–74.

29 Elchebly M, Payette P, Michaliszyn E, Crom-lish W, Collins S, Loy AL, et al. Increased in-sulin sensitivity and obesity resistance in mice lacking the protein tyrosine phosphatase-1B gene. Science. 1999 Mar; 283(5407): 1544–8.

30 Zabolotny JM, Bence-Hanulec KK, Stricker-Krongrad A, Haj F, Wang Y, Minokoshi Y, et al. PTP1B regulates leptin signal transduction in vivo. Dev Cell. 2002 Apr; 2(4): 489–95.

31 Zabolotny JM, Kim YB, Welsh LA, Kershaw EE, Neel BG, Kahn BB. Protein-tyrosine phosphatase 1B expression is induced by in-flammation in vivo. J Biol Chem. 2008 May;

283(21): 14230–41.32 Wu Y, Ouyang JP, Wu K, Wang SS, Wen CY,

Xia ZY. Rosiglitazone ameliorates abnormal expression and activity of protein tyrosine phosphatase 1B in the skeletal muscle of fat-fed, streptozotocin-treated diabetic rats. Br J Pharmacol. 2005 Sep; 146(2): 234–43.

33 Dodd GT, Andrews ZB, Simonds SE, Michael NJ, DeVeer M, Brüning JC, et al. A Hypotha-lamic Phosphatase Switch Coordinates Ener-gy Expenditure with Feeding. Cell Metab. 2017 Aug; 26(2): 375–393.e7.

34 White CL, Whittington A, Barnes MJ, Wang Z, Bray GA, Morrison CD. HF diets increase hypothalamic PTP1B and induce leptin resis-tance through both leptin-dependent and -in-dependent mechanisms. Am J Physiol Endo-crinol Metab. 2009 Feb; 296(2):E291–9.

35 Han Y, Belley M, Bayly CI, Colucci J, Du-fresne C, Giroux A, et al. Discovery of [(3-bro-mo-7-cyano-2-naphthyl)(difluoro)methyl]phosphonic acid, a potent and orally active small molecule PTP1B inhibitor. Bioorg Med Chem Lett. 2008 Jun; 18(11): 3200–5.

36 Van Heek M, Compton DS, France CF, Te-desco RP, Fawzi AB, Graziano MP, et al. Diet-induced obese mice develop peripheral, but not central, resistance to leptin. J Clin Invest. 1997 Feb; 99(3): 385–90.

37 El-Haschimi K, Pierroz DD, Hileman SM, Bjørbaek C, Flier JS. Two defects contribute to hypothalamic leptin resistance in mice with diet-induced obesity. J Clin Invest. 2000 Jun;

105(12): 1827–32.38 Münzberg H, Myers MG Jr. Molecular and

anatomical determinants of central leptin re-sistance. Nat Neurosci. 2005 May; 8(5): 566–70.

39 Ernst MB, Wunderlich CM, Hess S, Paehler M, Mesaros A, Koralov SB, et al. Enhanced Stat3 activation in POMC neurons provokes negative feedback inhibition of leptin and in-sulin signaling in obesity. J Neurosci. 2009 Sep; 29(37): 11582–93.

40 Martin TL, Alquier T, Asakura K, Furukawa N, Preitner F, Kahn BB. Diet-induced obesity alters AMP kinase activity in hypothalamus and skeletal muscle. J Biol Chem. 2006 Jul;

281(28): 18933–41.41 Knight ZA, Hannan KS, Greenberg ML,

Friedman JM. Hyperleptinemia is required for the development of leptin resistance. PLoS One. 2010 Jun; 5(6):e11376.

42 Belgardt BF, Brüning JC. CNS leptin and in-sulin action in the control of energy homeo-stasis. Ann N Y Acad Sci. 2010 Nov; 1212(1):

97–113.43 Garcia-Galiano D, Borges BC, Donato J Jr, Al-

len SJ, Bellefontaine N, Wang M, et al. PI3Kα inactivation in leptin receptor cells increases leptin sensitivity but disrupts growth and re-production. JCI Insight. 2017 Dec; 2(23):

96728.44 Woods SC, Lotter EC, McKay LD, Porte D Jr.

Chronic intracerebroventricular infusion of insulin reduces food intake and body weight of baboons. Nature. 1979 Nov; 282(5738):

503–5.45 Dodd GT, Decherf S, Loh K, Simonds SE,

Wiede F, Balland E, et al. Leptin and insulin act on POMC neurons to promote the brown-ing of white fat. Cell. 2015 Jan; 160(1–2): 88–104.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M


46 Obici S, Zhang BB, Karkanias G, Rossetti L. Hypothalamic insulin signaling is required for inhibition of glucose production. Nat Med. 2002 Dec; 8(12): 1376–82.

47 Pocai A, Morgan K, Buettner C, Gutierrez-Juarez R, Obici S, Rossetti L. Central leptin acutely reverses diet-induced hepatic insulin resistance. Diabetes. 2005 Nov; 54(11): 3182–9.

48 Hill JW, Elias CF, Fukuda M, Williams KW, Berglund ED, Holland WL, et al. Direct insu-lin and leptin action on pro-opiomelanocor-tin neurons is required for normal glucose ho-meostasis and fertility. Cell Metab. 2010 Apr;

11(4): 286–97.49 Fujikawa T, Berglund ED, Patel VR, Rama-

dori G, Vianna CR, Vong L, et al. Leptin en-gages a hypothalamic neurocircuitry to per-mit survival in the absence of insulin. Cell Metab. 2013 Sep; 18(3): 431–44.

50 Heni M, Kullmann S, Preissl H, Fritsche A, Häring HU. Impaired insulin action in the human brain: causes and metabolic conse-quences. Nat Rev Endocrinol. 2015 Dec;

11(12): 701–11.51 Heni M, Wagner R, Kullmann S, Veit R, Mat

Husin H, Linder K, et al. Central insulin ad-ministration improves whole-body insulin sensitivity via hypothalamus and parasympa-thetic outputs in men. Diabetes. 2014 Dec;

63(12): 4083–8.52 Vikram A, Jena G. S961, an insulin receptor

antagonist causes hyperinsulinemia, insulin-resistance and depletion of energy stores in rats. Biochem Biophys Res Commun. 2010 Jul; 398(2): 260–5.

53 Plum L, Schubert M, Brüning JC. The role of insulin receptor signaling in the brain. Trends Endocrinol Metab. 2005 Mar; 16(2): 59–65.

54 Niswender KD, Schwartz MW. Insulin and leptin revisited: adiposity signals with over-lapping physiological and intracellular signal-ing capabilities. Front Neuroendocrinol. 2003 Jan; 24(1): 1–10.

55 Urayama A, Banks WA. Starvation and tri-glycerides reverse the obesity-induced im-pairment of insulin transport at the blood-brain barrier. Endocrinology. 2008 Jul; 149(7):

3592–7.56 Kaiyala KJ, Prigeon RL, Kahn SE, Woods SC,

Schwartz MW. Obesity induced by a high-fat diet is associated with reduced brain insulin transport in dogs. Diabetes. 2000 Sep; 49(9):

1525–33.57 Bjørbaek C, Elmquist JK, Frantz JD, Shoelson

SE, Flier JS. Identification of SOCS-3 as a po-tential mediator of central leptin resistance. Mol Cell. 1998 Mar; 1(4): 619–25.

58 Münzberg H, Flier JS, Bjørbaek C. Region-specific leptin resistance within the hypothal-amus of diet-induced obese mice. Endocri-nology. 2004 Nov; 145(11): 4880–9.

59 Klaman LD, Boss O, Peroni OD, Kim JK, Martino JL, Zabolotny JM, et al. Increased en-ergy expenditure, decreased adiposity, and tissue-specific insulin sensitivity in protein-tyrosine phosphatase 1B-deficient mice. Mol Cell Biol. 2000 Aug; 20(15): 5479–89.

60 Loh K, Fukushima A, Zhang X, Galic S, Briggs D, Enriori PJ, et al. Elevated hypothalamic TCPTP in obesity contributes to cellular leptin resistance. Cell Metab. 2011 Nov; 14(5):

684–99.61 Pedroso JA, Buonfiglio DC, Cardinali LI, Fu-

rigo IC, Ramos-Lobo AM, Tirapegui J, et al. Inactivation of SOCS3 in leptin receptor-ex-pressing cells protects mice from diet-in-duced insulin resistance but does not prevent obesity. Mol Metab. 2014 Jun; 3(6): 608–18.

62 Tsou RC, Zimmer DJ, De Jonghe BC, Bence KK. Deficiency of PTP1B in leptin receptor-expressing neurons leads to decreased body weight and adiposity in mice. Endocrinology. 2012 Sep; 153(9): 4227–37.

63 Balland E, Chen W, Dodd GT, Conductier G, Coppari R, Tiganis T, et al. Leptin Signaling in the Arcuate Nucleus Reduces Insulin's Ca-pacity to Suppress Hepatic Glucose Produc-tion in Obese Mice. Cell Rep. 2019 Jan 8;

26(2): 346–355.e3.64 Lam BY, Cimino I, Polex-Wolf J, Nicole

Kohnke S, Rimmington D, Iyemere V, et al. Heterogeneity of hypothalamic pro-opiomel-anocortin-expressing neurons revealed by single-cell RNA sequencing. Mol Metab. 2017 Mar; 6(5): 383–92.

65 Williams KW, Margatho LO, Lee CE, Choi M, Lee S, Scott MM, et al. Segregation of acute leptin and insulin effects in distinct popula-tions of arcuate proopiomelanocortin neu-rons. J Neurosci. 2010 Feb; 30(7): 2472–9.

Dow

nloa

ded

by:

Mon

ash

Uni

vers

ity13

0.19

4.14

6.14

0 -

11/2

9/20

19 4

:52:

00 A

M

persistent leptin signaling in the arcuate nucleus impairs ... · lating leptin, or...

Documents